Test of Cochran-Armitage Continuity Correction to Test Trend on Proportion Data

Abstract

A nonparametric approach to analyzing trends in binary responses is widely applied in many fields of research. To test whether there is an increasing or decreasing trend can use the proportion data based on the Cochran-Armitage statistical test. The Cochran-Armitage test is a sensitive test on the score, the score here is 1 represents the 'success' event and 0 represents the 'fail' event. If the determination of the number of scores behind the position should not change the results of conclusions but in practice obtained different conclusions. To solve the problem, a correction of the Cochran-Armitage statistic called the Cochran-Armitage contingency correction test was performed. Cochran-Armitage continuity correction test will be applied to data of child mortality rate in Palembang City and South Sumatera in 2007-2015, which aims to determine whether there is an increase or decrease based on data of proportion on AKA. The result of the analysis shows that the test statistic value of Cochran-Armitage on AKA SUMSEL stimulus is -1,27414 it shows that from 2007-2015 the proportion of child mortality rate has the same trend, with indication of continuity correction equal to 2,84340 means from 2007 -2015 the proportion of child mortality rate in Palembang City has a tendency of declining trend. Meanwhile, the time series data stimulus obtained by the value of test statistic at Cochran-Armitage of 11,96437 it shows that from 2007-2015 the proportion of child mortality has a tendency of declining trend, with indigo correction continuity of 4 means from 2007-2015 The proportion of child mortality in each year there is no trend.